Counterfeit AKM Parts

[gnuarm]

I was bitten by the counterfeit chip problem.  Parts off Alibaba did not work in my design while a previous 8,000 genuine parts did.  Here is a good part with important features noted.

Here are the counterfeit chips with the features marked.

The lettering style is different, but it seems AKM uses different markings at different assembly plants.  Here is one I sent to AKM some time ago when we suspected some devices.

The marking on the suspect chip and the previously suspected chip are similar, but the previously suspected chip turned out to work ok.  The thing in common with the good chips is the pin 1 mark is molded rather than laser etched as on the suspected chip. 

I've had some back and forth with the supplier.  Initially he wanted me to test more chips, but I explained the contract assembler is 200 miles from me and they charge for their time, so it would be $1000 for the trip and expenses.   He's talking to his supplier.

I don't think this guy knew the parts were counterfeit.  His source might not know.  What I'm wondering now is how many of these parts are like this?  Each vendor has a different price.  Everyone I tell about there being a lower price says anyone cheaper than them are not selling genuine parts.  So maybe that means they are all fakes.  That's the really bad part of this.  I don't know if I can get real parts at all. 

[Bud]

People who buy parts for production from alibaba must be living in a parallel Universe....

[wraper]

People who buy parts for production from alibaba must be living in a parallel Universe....

FYI AKM factory burned down and if you want to produce anything without redesigning, you will have no other option other than to buy at shady places.

[gnuarm]

I'm expecting a $4 million order and it would be nice to be able to fill it.  I believe there are still parts available.  The issue is what you have to pay to get them them.  Apparently something more than what I paid for these.

[AndyC_772]

If you're expecting a $4M order, that $1000 assembly / test fee is lost in the rounding errors. Get on and pay it, or buy a heat gun and swap the parts yourself, it's a total no-brainer.

[gnuarm]

I don't think you understand.  I've already tested the parts.  They are counterfeit.  Done and dusted.  It's the seller who was asking about testing more units.  What are the chances that you pick two out of 500 (I used one from each end of the tape so they weren't next to each other) and they are both bad if the parts are real?  I don't think we've had any bad devices of his part number in building 8,000 units.  Even if the parts were 1 bad in 1000 parts on the average, seeing two back to back is a 1 in 1,000,000 chance. 

Maybe I should forget electronics and play the lottery or bet at the race track?

[phil from seattle]

Maybe I should forget electronics and play the lottery or bet at the race track?

Based on your record luck with AKM parts, I'd say stick to electronics...

But...

Part of me says any parts that don't come from licensed distirbutors are likely to be suspect.  Any parts that come from a China based middleman are probably fakes. You are between a rock and a hard place. I think the suggestion to redesign is your best bet.  I am skeptical that you can test your way out of it.

One thought is to talk directly to AKM - maybe they can point you in the right direction.  Like a large buyer of the part.  If, indeed, the parts are actually out there, you need to follow the distribution chain to find who might be holding them.

And, I would go back to your customer and renegotiate the deal.  That happens all the time.  If they want your product, they may be willing to wait longer or pay more for sooner delivery.

good luck!

[bdunham7]

If I understand correctly this is the deal you posted previously where you have a contractor who resells to the government or DOD or something like that and you wanted to avoid redesign and felt that you could pass almost any cost onto your customer.

So two three questions.

1)  How much is it worth to not redesign to an available part?  Is it worth most of a million dollars to avoid that--and still have to deal with a sketchy supply chain and a part that will likely never come back to production?  What happens in a few years when they need another batch?

2)  What happens if the parts are counterfeit but they pass testing, only to have a higher failure rate or some other problem down the line?  Is testing really an acceptable substitute for documented supply chain management?

3) Have you consulted an attorney to determine the scope of your potential liabilities in this?  And I don't just mean civil liabilities.

My unsolicited, uncompensated legal advice is......"Yikes, dude!"

[gnuarm]

If I understand correctly this is the deal you posted previously where you have a contractor who resells to the government or DOD or something like that and you wanted to avoid redesign and felt that you could pass almost any cost onto your customer.

That's close.  I'm impressed you remember all that.  I sell my board to a major networking company who uses it in their systems which are sold to the general market which is often the federal government. 

So two three questions.

1)  How much is it worth to not redesign to an available part?  Is it worth most of a million dollars to avoid that--and still have to deal with a sketchy supply chain and a part that will likely never come back to production?  What happens in a few years when they need another batch?

This is all stuff that is in consideration, but not the immediate issue.  They are projecting a need this summer for some number of units (It's always fuzzy to me if units are *my* units or theirs as there can be a 2:1 or even 8:1 difference depending on the product).  So a new design could not be ready before then since it would require various certs. 

I contacted them soon after the factory fire and being a large company it is taking them a long time to weigh the factors.  Right now they are talking about buying some number of my boards for the 2 year forecast.  Anything beyond that would be a new design which is in consideration. 

2)  What happens if the parts are counterfeit but they pass testing, only to have a higher failure rate or some other problem down the line?  Is testing really an acceptable substitute for documented supply chain management?

What happens if any part on the board in spite of "documented supply chain management" develops a high failure rate?  I provide a year warranty and beyond that its on the customer.  I don't think the counterfeits are real parts.  They often remark a different part in the same package, so the parts don't work from day one. 

3) Have you consulted an attorney to determine the scope of your potential liabilities in this?  And I don't just mean civil liabilities.

My unsolicited, uncompensated legal advice is......"Yikes, dude!"

Yikes, indeed!

[bdunham7]

What happens if any part on the board in spite of "documented supply chain management" develops a high failure rate?

Have your lawyer explain the difference between your product developing a high failure rate and causing mayhem due to failures of genuine parts that you procured through documented channels and approved suppliers vs due to the failures of some parts you bought from some shady vendor in an alley (like Alley Baba) at a cut rate price. Of course it depends upon the consequences of the failure to some degree, but those are two different scenarios entirely. 

[NorthGuy]

It says 4556VTP. According to the datasheet, it should be marked as 4556VT. It would be silly for the counterfeiters to add extra letters. Might be some special part, or a different part altogether. I would verify this with the manufacturer.

[gnuarm]

It says 4556VTP. According to the datasheet, it should be marked as 4556VT. It would be silly for the counterfeiters to add extra letters. Might be some special part, or a different part altogether. I would verify this with the manufacturer.

Good catch, but it is a marking used internally. 

From the AKM web site... one of the few useful pages there.  Their site has become uniquely ineffective.   I have an account on the web site but there is no way to log in.   

[Note]
You may sometimes see "P" or "-8" character in AKM part number. This type of the part number is internally used and it is the same product as the one without these characters.

Example: "AK□□□□-8" and "AK□□□□" are the same products.

[SilverSolder]

[...]  Alley Baba [...]

 

[SMTech]

Other people will have this chip in their designs, some of them will have more reason to resist redesign than you (e.g directly into military/aerospace/automotive) that has an extensive test and verification process they can't avoid or it was already ordered years ago. Others will have had deeper pockets and quicker responses and purchased every bit of stock they could find. I think you need a really really good reason not to design in another chip, while many of the others DACs out there do have config over SPI/I2C/whatever you wanted to avoid, many also have some config that can be done without or at least a reasonable default, for a contract of that value I'd say it was worth it to have an alternative variant spun up around something sourceable. At the very least I think it would be an idea to let a specialist company who already has the contacts and facilities to track down verify and test possible stock, at least that way if something goes badly you can share the blame.

Some factories run parts at fixed schedules, if the fire caught AKM right at the wrong point that stock you think exists may not, I'd be very concerned if none of the typical only slightly shady brokers claim to have access to stock because that means even they don't trust the vendors that claim to have it.

While I'm sure you've tried it. Putting that part into some brokers sites lists 9000 where the manufacturer field is blank (laziness?) and 2500 under AKM I can see that on https://www.electronics-direct.com/ who are English and I know and some random US site https://www.broadlinecomponents.com/ so clearly pulling that data from the same place. Octopart find lower alleged volumes via two other places who are less clear on their status although they look like brokers to me.

I don't think it matters if you warranty something for a year and slightly dodgy parts just eke you out past that, if the end customer starts seeing a beyond reasonable failure rate, they could investigate it themselves and start following the traceability trail. Given you live in the land of suing the socks off everyone in sight, I wouldn't want to be the holder of that broken link.

[rsjsouza]

I am not surprised that, after such event, the market is flooded with counterfeit parts.
 

I contacted them soon after the factory fire and being a large company it is taking them a long time to weigh the factors.  Right now they are talking about buying some number of my boards for the 2 year forecast.  Anything beyond that would be a new design which is in consideration. 

In the similar scenarios I have experienced, unfortunately at this point there's not much to be done apart from contacting the manufacturer - after all, this event is six months old and maybe they have an alternative. Back when you contacted them, their procurement department should have been nimble... I have seen this scenario multiple times and was at the other end of the phone with mad customers unable to fulfill production deadlines.

[Mangozac]

What happens if any part on the board in spite of "documented supply chain management" develops a high failure rate?

Have your lawyer explain the difference between your product developing a high failure rate and causing mayhem due to failures of genuine parts that you procured through documented channels and approved suppliers vs due to the failures of some parts you bought from some shady vendor in an alley (like Alley Baba) at a cut rate price. Of course it depends upon the consequences of the failure to some degree, but those are two different scenarios entirely.

This - you can claim whatever warranty you want but if you are caught out knowingly sourcing very questionable parts (and what's more, documenting it on an open forum like this!) you're opening yourself up to litigation no matter what country you're in!

I would have commenced redesign the instant I became aware of the issue. As much as a headache a redesign is, to me it's far less of a headache than sourcing an obscure part from random Asian suppliers!

[jonpaul]

Our associated firm makes cinema/pro audio devices. After exisitng stock is depleted, we   halted production of all AKM containing  products in 2021.
Redesign is one option.

The many IC fab plant fires are suspicious, Renases, AKM, Taiwan.

Perhaps we have IC wars?

Jon

[phil from seattle]

Our associated firm makes cinema/pro audio devices. After exisitng stock is depleted, we   halted production of all AKM containing  products in 2021.
Redesign is one option.

The many IC fab plant fires are suspicious, Renases, AKM, Taiwan.

Perhaps we have IC wars?

Jon

i think it is FREE insurance $$$

Not if you get caught.

[wraper]

i think it is FREE insurance $$$

First of all, insurance companies do not like to give their money away and will try to find any loophole possible to not pay. Secondly, why would you chose the worst moment possible when factories are churning at maximum load possible due to component shortage.

[gnuarm]

i think it is FREE insurance $$$

First of all, insurance companies do not like to give their money away and will try to find any loophole possible to not pay. Secondly, why would you chose the worst moment possible when factories are churning at maximum load possible due to component shortage.

He's just messing.  The company was very seriously hurt by the fire.  They lost a lot of money in this fire directly.  Insurance only covers assets.  They have many employees who are out of work, so find work elsewhere losing their experience.  They have lost a lot of sales while the plant is down.  Worse, it is too expensive to resume manufacturing on many of their older products which were still selling well costing them profits indefinitely. 

The part I need is one of these parts they are not putting back into production.  The part they are recommending for replacement is 24 pins vs. 20 for the old one.  The package is smaller being a QFN, but for no good reason has a thermal pad eating up more real estate than the TSSOP it replaces. 

The real issue for me is once the respin can of worms is open, the FPGA gets replaced and there are no good alternatives.  Everything available is either long in the tooth or uses a BGA package, mostly very difficult to route BGAs with way more pins than needed.  Lattice has a number of QFN/QFP package options, but not with enough I/Os (~60). 

I have often lamented how the FPGA world is so focused on the high dollar devices.  Lattice supports the mobile market with some small parts, but with very tiny packages that are not practical for my work.  I may end up replacing a 100 pin QFP with a 256 pin BGA. 

[jonpaul]

Our colleagues  took about 1 yr to redesign old audio products to replace costly and obscelescent DSPs from Motorola or TI, with ARM.



Jon

[SilverSolder]

i think it is FREE insurance $$$

First of all, insurance companies do not like to give their money away and will try to find any loophole possible to not pay. Secondly, why would you chose the worst moment possible when factories are churning at maximum load possible due to component shortage.

He's just messing.  The company was very seriously hurt by the fire.  They lost a lot of money in this fire directly.  Insurance only covers assets.  They have many employees who are out of work, so find work elsewhere losing their experience.  They have lost a lot of sales while the plant is down.  Worse, it is too expensive to resume manufacturing on many of their older products which were still selling well costing them profits indefinitely. 

The part I need is one of these parts they are not putting back into production.  The part they are recommending for replacement is 24 pins vs. 20 for the old one.  The package is smaller being a QFN, but for no good reason has a thermal pad eating up more real estate than the TSSOP it replaces. 

The real issue for me is once the respin can of worms is open, the FPGA gets replaced and there are no good alternatives.  Everything available is either long in the tooth or uses a BGA package, mostly very difficult to route BGAs with way more pins than needed.  Lattice has a number of QFN/QFP package options, but not with enough I/Os (~60). 

I have often lamented how the FPGA world is so focused on the high dollar devices.  Lattice supports the mobile market with some small parts, but with very tiny packages that are not practical for my work.  I may end up replacing a 100 pin QFP with a 256 pin BGA.

Is it an option to use multiple of the tiny FPGAs in parallel, or don't things work that tidily?

[gnuarm]

Is it an option to use multiple of the tiny FPGAs in parallel, or don't things work that tidily?

Anything is possible, but it would be a large PITA.  Even if possible, multiple small devices end up needing more total pins and more real estate which is the big problem. 

I've never worked on an FPGA DAC, but it might be possible to use sigma delta converters within the FPGA or even PWM.  I'm presently working on FPGA based ADC for a project with much less demanding requirements.  16 bits at 48 ksps requires over 3 Gbps.  I don't think that is realistic.  Telephony quality is much easier, 13/14 bits at 8 ksps requiring 131 MHz.

[thm_w]

I don't know if I can get real parts at all.

I'm certain you can, its just going to be a large effort. eg those "used" parts we found in the previous thread: https://www.eevblog.com/forum/manufacture/buying-on-alibaba/
You'd likely have to factor in the cost of someone either creating a test jig, or, hand re-working any boards that failed initial testing. Having a CM deal with possible counterfeit parts is a large time/money sink.

The funny thing is you were going to use parts that were clearly sanded + re-lasered. So they may have been salvaged (if they worked).

[gnuarm]

No, there was no "going to use parts that were sanded...".  The markings are not obviously wrong until we had the failures, then we looked at them harder.  It was only under the microscope that they appeared to be odd.  The pin 1 marker is a good indicator although that could be made correctly. 

When I compared the parts under the microscope I thought the differences indicated counterfeit parts.  But I had gone through this exercise previously when another fab house had trouble getting boards to work.  Looking at the images the suspect part had very similar markings to the counterfeit chip in this case.  However, AKM confirmed that chip was good.  In this case the only visual indicators was the laser marked pin 1 indicator and after contacting AKM for these parts, they confirmed the date code is invalid.